Figure 1: Pacific Giant Salamander (Dicamptodon tenebrosus), in Opal Creek Wilderness, Oregon. Photographer: Dave Huth.


Introduction

Pacific giant salamanders (Dicamptodon tenebrosus), the largest terrestrial salamander in North America, thrive in the wet forests of coastal northern California, Oregon, and Washington (California Herps, 2019). Their preferred habitat is near clear, cold streams, rivers, lakes, or ponds with rocks, logs or root channels to burrow under. Larvae hatch under water and spend approximately the first two years of their lives in the water. Once they transform to their terrestrial forms, reproduction also occurs in the water, (California Herps, 2019). Due to their sensitivity to water quality, Pacific giant salamanders are used as an indicator species of the declining health of waterways. Logging can be particularly impactful on Pacific giant salamanders due to the increased sedimentation of streams from erosion (Oregon Wild, 2019).

Oregon State University has been monitoring the species along Mack Creek in the H.J. Andrews Experimental Forest near Eugene, Oregon since 1993. Electrofishing is used to capture vertebrates along the creek and all captured vertebrates are measured and weighed (Gregory S. V., 2016). The Mack Creek watershed (1,433 acres) contains two unique management histories: approximately 13% of the watershed was clear cut around 1963 and the rest contains a 500-year-old coniferous forest (Andrews Experimental Forest, 2017). This research forest offers a unique chance to consider the long-term impacts of logging on salamander well-being compared to a more natural forest.

In this analysis, we use data provided by the Andrews Experimental Forest to explore differences in salamander abundance and weight among old growth and clear-cut sections of Mack Creek and among different stream characteristics (pool, cascade, side channel).


Data & methods

Figure 2: Map of HJ Andrews Forest Experimental Watersheds - Mack Creek watershed highlighted in green. Source: HJ Andrews Experimental Forest.


Data for populations of Pacific giant salamanders (Dicamptodon tenebrosus) were collected in the H.J. Andrews Experimental Forest near Blue River, Oregon as part of the National Science Foundation’s Long Term Ecological Research Network. Salamanders were sampled with two-pass electrofishing in two sections of Mack Creek, one reach in a section of clearcut forest (ca. 1963) and the other an upstream 500-year-old coniferous forest. The two sections are then subdivided to indicate where in the channel salamanders were found (pool, cascade, or side-channel). According to project metadata, the purpose of the Aquatic Vertebrate Population Study in Mack Creek, is “to provide a long-term record of populations in aquatic vertebrates…related to land use and disturbance histories.”

For more information see the project metadata

The following visual analyses and statistical tests were performed in this analysis to explore Pacific giant salamander populations in Mack Creek: - Visual comparion of salamander counts in old growth and clear cut sections of Mack Creek; - Chi-square test to test whether where in the channel salamanders are found varies based on forest section (clear cut vs. old growth); - Two-sided, unpaired t-test to see if there is a significant difference in salander weights between clear cut and old growth forest sections; - And, to examine differences in mean salamander weights in 2017 between channel classifications, we performed one-way ANOVA, Tukey’s HSD, and Cohen’s effect size tests.

All tests were performed using a significance level of \(\alpha\) = .05. All analyses and figures were prepared using R software version 3.6.1.


Results

1. Salamander abundance in old-growth and clear-cut sections of Mack Creek

Figure 3 demonstrates that lobster abundance has followed relatively similar trends in Mack Creek depite the different forest management techniques. From 1995 to 2011, salamander counts along the clear-cut sections of Mack Creek were lower than those in the old-growth sections. In the years preceding and following this range, there were years where counts in the clear-cut sections exceeded those in the old-growth sections. Overall trends in abundance seem to impact salamanders no matter what forest type they are in, with both groups experiencing increading and decreasing numbers in the same years.

Figure 3: Salamander abundance in clear-cut and old-growth forest, 1993 to 2017. Annual salamander counts through electro-fishing in Mack Creek in the HJ Andrews Experimental Forest. Interactive: Double-click on individual site within legend to highlight. Single-click and unclick sites within legend to toggle visibility and compare site data.


2. Salamander counts by channel classification, 2017

After exploring salamander abundance based on forest section, we wanted to explore whether salamander abundance in different parts of Mack Creek is related to which forest section they are found in. The contingency table below, which we use in part 3 to conduct a Chi-square test, shows counts and proportions of where salamanders were found in the channel in both the clear cut and old growth sections of the H.J. Andrews Experimental Forest.

Table 1: 2017 counts of Salamanders found in Mack Creek, grouped by section (clear cut vs. old growth) and position within channel (pool, cascades, side-channel).

Section Cascades Pool Side channel
Clear Cut 247 (67%) 31 (8%) 90 (24%)
Old Growth 201 (63%) 45 (14%) 74 (23%)


3. Association between forest section and channel classification, 2017

We used a Chi-square test to examine the relationship between channel position (pool, side-channel or cascades) and section (clear cut or old growth) in Mack Creek. We found a \(\chi\)2 value of 5.54 (d.f. 2) with a p-value of 0.063. As a result, we retain the null hypothesis that there is no association between channel position and forest section for Pacific giant salamanders in Mack Creek.


4. Salamander weights in clear cut vs. old growth forests, 2017

In order to assess whether forest management strategies had an impact on salamander weight and abundance, statistical testing (Student’s T Test and Cohen’s Effect Size) was used to compare salamander weights in clear cut and old growth forest sections in 2017.

In 2017, there was a 17% difference between salamander weights at clear-cut forest (m = 7.78 grams) and old-growth forest (m = 6.58 grams) sections. This difference is not statistically significant (t(692.79) = 1.67, p = 0.096), with a negligible effect size (0.13).


5. Comparison of salamander weights by channel classification

Figure 4: Comparison of salamander weights in three different channel classifications, 2017. Salamander weights (grams) measured in side-channels, pools, and cascades of Mack Creek in the HJ Andrews Experimental Forest. Mean salamander weight shown by the small, black square with black whiskers showing standard error.

While Figure 4 displays means and variance for Salamander weights in Mack Creek based on channel classification, the below table gives a more robust summary of the data.

Table 2: Summary statistics for weights of Salamanders found in Mack Creek, grouped by position within channel (pool, cascades, side-channel), in 2017.
Channel position Mean weight (g) Median weight (g) Standard deviation Sample size Standard error Variance
Cascades 7.52 5.37 9.03 448 0.43 81.50
Pool 9.30 5.63 13.62 76 1.56 185.58
Side-Channel 5.68 3.79 8.27 164 0.65 68.31

Mean weight (g) differed significantly between salamanders found in (one-way ANOVA with post-hoc Tukey’s HSD, (F(2, 684) = 4.22, p = 0.015).

An analysis of variance (ANOVA) on salamander weights in Mack Creek yielded significant variation among channel positions, (F(2, 684) = 4.22, p = 0.015). A post hoc Tukey test showed that mean weights for salamanders found in side-channels and pools differed significantly p = 0.017) with a small effect size (0.35) and a 48% difference. We do not see siginificant differences elsewhere.

While the above analysis uses mean as the measure of central tendency, there are numerous outliers in the data for all three channel classifications. This is easily visible in Figure 4. As a result, means of these samples will skew greater than other measures of central tendency. Therefore we present sample medians for all three channel classifications. Aligning with the results from statistical tests (one-way ANOVA and post-hoc Tukey’s HSD) to compare means, we can see that side-channel median weight (median = 3.79) is lowest. Overall, Pacific giant salamander weights are relatively similar between the three channel classifications in Mack Creek.


Summary


References

Andrews Experimental Forest. 2017. Location Name: Andrews Mack Creek Watershed. HJ Andrews Experimental Forest Long-Term Ecological Research.

California Herps. 2019. Coastal Giant Salamander - Dicamptodon tenebrosus.

Gregory S. V. 2016. Aquatic Vertebrate Population Study in Mack Creek, Andrews Experimental Forest, 1987 to present. Environmental Data Initiative. Dataset accessed 11/19/2019.

Oregon Wild. 2019. Pacific Giant Salamander.